xref: /dflybsd-src/usr.bin/tr/cmap.c (revision fbc9049bb05b3ca966e87d4c55eb360cbc38f5d3)
1*fbc9049bSSascha Wildner /*-
2*fbc9049bSSascha Wildner  * Copyright (c) 2004 Tim J. Robbins.
3*fbc9049bSSascha Wildner  * All rights reserved.
4*fbc9049bSSascha Wildner  *
5*fbc9049bSSascha Wildner  * Redistribution and use in source and binary forms, with or without
6*fbc9049bSSascha Wildner  * modification, are permitted provided that the following conditions
7*fbc9049bSSascha Wildner  * are met:
8*fbc9049bSSascha Wildner  * 1. Redistributions of source code must retain the above copyright
9*fbc9049bSSascha Wildner  *    notice, this list of conditions and the following disclaimer.
10*fbc9049bSSascha Wildner  * 2. Redistributions in binary form must reproduce the above copyright
11*fbc9049bSSascha Wildner  *    notice, this list of conditions and the following disclaimer in the
12*fbc9049bSSascha Wildner  *    documentation and/or other materials provided with the distribution.
13*fbc9049bSSascha Wildner  *
14*fbc9049bSSascha Wildner  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15*fbc9049bSSascha Wildner  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16*fbc9049bSSascha Wildner  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17*fbc9049bSSascha Wildner  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18*fbc9049bSSascha Wildner  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19*fbc9049bSSascha Wildner  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20*fbc9049bSSascha Wildner  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21*fbc9049bSSascha Wildner  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22*fbc9049bSSascha Wildner  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23*fbc9049bSSascha Wildner  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24*fbc9049bSSascha Wildner  * SUCH DAMAGE.
25*fbc9049bSSascha Wildner  *
26*fbc9049bSSascha Wildner  * $FreeBSD: head/usr.bin/tr/cmap.c 200462 2009-12-13 03:14:06Z delphij $
27*fbc9049bSSascha Wildner  */
28*fbc9049bSSascha Wildner /*
29*fbc9049bSSascha Wildner  * "Character map" ADT. Stores mappings between pairs of characters in a
30*fbc9049bSSascha Wildner  * splay tree, with a lookup table cache to simplify looking up the first
31*fbc9049bSSascha Wildner  * bunch of characters (which are presumably more common than others).
32*fbc9049bSSascha Wildner  */
33*fbc9049bSSascha Wildner 
34*fbc9049bSSascha Wildner #include <assert.h>
35*fbc9049bSSascha Wildner #include <limits.h>
36*fbc9049bSSascha Wildner #include <stdbool.h>
37*fbc9049bSSascha Wildner #include <stdlib.h>
38*fbc9049bSSascha Wildner #include <wchar.h>
39*fbc9049bSSascha Wildner #include "cmap.h"
40*fbc9049bSSascha Wildner 
41*fbc9049bSSascha Wildner static struct cmapnode *cmap_splay(struct cmapnode *, wint_t);
42*fbc9049bSSascha Wildner 
43*fbc9049bSSascha Wildner /*
44*fbc9049bSSascha Wildner  * cmap_alloc --
45*fbc9049bSSascha Wildner  *	Allocate a character map.
46*fbc9049bSSascha Wildner  */
47*fbc9049bSSascha Wildner struct cmap *
cmap_alloc(void)48*fbc9049bSSascha Wildner cmap_alloc(void)
49*fbc9049bSSascha Wildner {
50*fbc9049bSSascha Wildner 	struct cmap *cm;
51*fbc9049bSSascha Wildner 
52*fbc9049bSSascha Wildner 	cm = malloc(sizeof(*cm));
53*fbc9049bSSascha Wildner 	if (cm == NULL)
54*fbc9049bSSascha Wildner 		return (NULL);
55*fbc9049bSSascha Wildner 	cm->cm_root = NULL;
56*fbc9049bSSascha Wildner 	cm->cm_def = CM_DEF_SELF;
57*fbc9049bSSascha Wildner 	cm->cm_havecache = false;
58*fbc9049bSSascha Wildner 	cm->cm_min = cm->cm_max = 0;
59*fbc9049bSSascha Wildner 	return (cm);
60*fbc9049bSSascha Wildner }
61*fbc9049bSSascha Wildner 
62*fbc9049bSSascha Wildner /*
63*fbc9049bSSascha Wildner  * cmap_add --
64*fbc9049bSSascha Wildner  *	Add a mapping from "from" to "to" to the map.
65*fbc9049bSSascha Wildner  */
66*fbc9049bSSascha Wildner bool
cmap_add(struct cmap * cm,wint_t from,wint_t to)67*fbc9049bSSascha Wildner cmap_add(struct cmap *cm, wint_t from, wint_t to)
68*fbc9049bSSascha Wildner {
69*fbc9049bSSascha Wildner 	struct cmapnode *cmn, *ncmn;
70*fbc9049bSSascha Wildner 
71*fbc9049bSSascha Wildner 	cm->cm_havecache = false;
72*fbc9049bSSascha Wildner 
73*fbc9049bSSascha Wildner 	if (cm->cm_root == NULL) {
74*fbc9049bSSascha Wildner 		cmn = malloc(sizeof(*cmn));
75*fbc9049bSSascha Wildner 		if (cmn == NULL)
76*fbc9049bSSascha Wildner 			return (false);
77*fbc9049bSSascha Wildner 		cmn->cmn_from = from;
78*fbc9049bSSascha Wildner 		cmn->cmn_to = to;
79*fbc9049bSSascha Wildner 		cmn->cmn_left = cmn->cmn_right = NULL;
80*fbc9049bSSascha Wildner 		cm->cm_root = cmn;
81*fbc9049bSSascha Wildner 		cm->cm_min = cm->cm_max = from;
82*fbc9049bSSascha Wildner 		return (true);
83*fbc9049bSSascha Wildner 	}
84*fbc9049bSSascha Wildner 
85*fbc9049bSSascha Wildner 	cmn = cm->cm_root = cmap_splay(cm->cm_root, from);
86*fbc9049bSSascha Wildner 
87*fbc9049bSSascha Wildner 	if (cmn->cmn_from == from) {
88*fbc9049bSSascha Wildner 		cmn->cmn_to = to;
89*fbc9049bSSascha Wildner 		return (true);
90*fbc9049bSSascha Wildner 	}
91*fbc9049bSSascha Wildner 
92*fbc9049bSSascha Wildner 	ncmn = malloc(sizeof(*ncmn));
93*fbc9049bSSascha Wildner 	if (ncmn == NULL)
94*fbc9049bSSascha Wildner 		return (false);
95*fbc9049bSSascha Wildner 	ncmn->cmn_from = from;
96*fbc9049bSSascha Wildner 	ncmn->cmn_to = to;
97*fbc9049bSSascha Wildner 	if (from < cmn->cmn_from) {
98*fbc9049bSSascha Wildner 		ncmn->cmn_left = cmn->cmn_left;
99*fbc9049bSSascha Wildner 		ncmn->cmn_right = cmn;
100*fbc9049bSSascha Wildner 		cmn->cmn_left = NULL;
101*fbc9049bSSascha Wildner 	} else {
102*fbc9049bSSascha Wildner 		ncmn->cmn_right = cmn->cmn_right;
103*fbc9049bSSascha Wildner 		ncmn->cmn_left = cmn;
104*fbc9049bSSascha Wildner 		cmn->cmn_right = NULL;
105*fbc9049bSSascha Wildner 	}
106*fbc9049bSSascha Wildner 	if (from < cm->cm_min)
107*fbc9049bSSascha Wildner 		cm->cm_min = from;
108*fbc9049bSSascha Wildner 	if (from > cm->cm_max)
109*fbc9049bSSascha Wildner 		cm->cm_max = from;
110*fbc9049bSSascha Wildner         cm->cm_root = ncmn;
111*fbc9049bSSascha Wildner 
112*fbc9049bSSascha Wildner 	return (true);
113*fbc9049bSSascha Wildner }
114*fbc9049bSSascha Wildner 
115*fbc9049bSSascha Wildner /*
116*fbc9049bSSascha Wildner  * cmap_lookup_hard --
117*fbc9049bSSascha Wildner  *	Look up the mapping for a character without using the cache.
118*fbc9049bSSascha Wildner  */
119*fbc9049bSSascha Wildner wint_t
cmap_lookup_hard(struct cmap * cm,wint_t ch)120*fbc9049bSSascha Wildner cmap_lookup_hard(struct cmap *cm, wint_t ch)
121*fbc9049bSSascha Wildner {
122*fbc9049bSSascha Wildner 
123*fbc9049bSSascha Wildner 	if (cm->cm_root != NULL) {
124*fbc9049bSSascha Wildner 		cm->cm_root = cmap_splay(cm->cm_root, ch);
125*fbc9049bSSascha Wildner 		if (cm->cm_root->cmn_from == ch)
126*fbc9049bSSascha Wildner 			return (cm->cm_root->cmn_to);
127*fbc9049bSSascha Wildner 	}
128*fbc9049bSSascha Wildner 	return (cm->cm_def == CM_DEF_SELF ? ch : cm->cm_def);
129*fbc9049bSSascha Wildner }
130*fbc9049bSSascha Wildner 
131*fbc9049bSSascha Wildner /*
132*fbc9049bSSascha Wildner  * cmap_cache --
133*fbc9049bSSascha Wildner  *	Update the cache.
134*fbc9049bSSascha Wildner  */
135*fbc9049bSSascha Wildner void
cmap_cache(struct cmap * cm)136*fbc9049bSSascha Wildner cmap_cache(struct cmap *cm)
137*fbc9049bSSascha Wildner {
138*fbc9049bSSascha Wildner 	wint_t ch;
139*fbc9049bSSascha Wildner 
140*fbc9049bSSascha Wildner 	for (ch = 0; ch < CM_CACHE_SIZE; ch++)
141*fbc9049bSSascha Wildner 		cm->cm_cache[ch] = cmap_lookup_hard(cm, ch);
142*fbc9049bSSascha Wildner 
143*fbc9049bSSascha Wildner 	cm->cm_havecache = true;
144*fbc9049bSSascha Wildner }
145*fbc9049bSSascha Wildner 
146*fbc9049bSSascha Wildner /*
147*fbc9049bSSascha Wildner  * cmap_default --
148*fbc9049bSSascha Wildner  *	Change the value that characters without mappings map to, and
149*fbc9049bSSascha Wildner  *	return the old value. The special character value CM_MAP_SELF
150*fbc9049bSSascha Wildner  *	means characters map to themselves.
151*fbc9049bSSascha Wildner  */
152*fbc9049bSSascha Wildner wint_t
cmap_default(struct cmap * cm,wint_t def)153*fbc9049bSSascha Wildner cmap_default(struct cmap *cm, wint_t def)
154*fbc9049bSSascha Wildner {
155*fbc9049bSSascha Wildner 	wint_t old;
156*fbc9049bSSascha Wildner 
157*fbc9049bSSascha Wildner 	old = cm->cm_def;
158*fbc9049bSSascha Wildner 	cm->cm_def = def;
159*fbc9049bSSascha Wildner 	cm->cm_havecache = false;
160*fbc9049bSSascha Wildner 	return (old);
161*fbc9049bSSascha Wildner }
162*fbc9049bSSascha Wildner 
163*fbc9049bSSascha Wildner static struct cmapnode *
cmap_splay(struct cmapnode * t,wint_t ch)164*fbc9049bSSascha Wildner cmap_splay(struct cmapnode *t, wint_t ch)
165*fbc9049bSSascha Wildner {
166*fbc9049bSSascha Wildner 	struct cmapnode N, *l, *r, *y;
167*fbc9049bSSascha Wildner 
168*fbc9049bSSascha Wildner 	/*
169*fbc9049bSSascha Wildner 	 * Based on public domain code from Sleator.
170*fbc9049bSSascha Wildner 	 */
171*fbc9049bSSascha Wildner 
172*fbc9049bSSascha Wildner 	assert(t != NULL);
173*fbc9049bSSascha Wildner 
174*fbc9049bSSascha Wildner 	N.cmn_left = N.cmn_right = NULL;
175*fbc9049bSSascha Wildner 	l = r = &N;
176*fbc9049bSSascha Wildner 	for (;;) {
177*fbc9049bSSascha Wildner 		if (ch < t->cmn_from) {
178*fbc9049bSSascha Wildner 			if (t->cmn_left != NULL &&
179*fbc9049bSSascha Wildner 			    ch < t->cmn_left->cmn_from) {
180*fbc9049bSSascha Wildner 				y = t->cmn_left;
181*fbc9049bSSascha Wildner 				t->cmn_left = y->cmn_right;
182*fbc9049bSSascha Wildner 				y->cmn_right = t;
183*fbc9049bSSascha Wildner 				t = y;
184*fbc9049bSSascha Wildner 			}
185*fbc9049bSSascha Wildner 			if (t->cmn_left == NULL)
186*fbc9049bSSascha Wildner 				break;
187*fbc9049bSSascha Wildner 			r->cmn_left = t;
188*fbc9049bSSascha Wildner 			r = t;
189*fbc9049bSSascha Wildner 			t = t->cmn_left;
190*fbc9049bSSascha Wildner 		} else if (ch > t->cmn_from) {
191*fbc9049bSSascha Wildner 			if (t->cmn_right != NULL &&
192*fbc9049bSSascha Wildner 			    ch > t->cmn_right->cmn_from) {
193*fbc9049bSSascha Wildner 				y = t->cmn_right;
194*fbc9049bSSascha Wildner 				t->cmn_right = y->cmn_left;
195*fbc9049bSSascha Wildner 				y->cmn_left = t;
196*fbc9049bSSascha Wildner 				t = y;
197*fbc9049bSSascha Wildner 			}
198*fbc9049bSSascha Wildner 			if (t->cmn_right == NULL)
199*fbc9049bSSascha Wildner 				break;
200*fbc9049bSSascha Wildner 			l->cmn_right = t;
201*fbc9049bSSascha Wildner 			l = t;
202*fbc9049bSSascha Wildner 			t = t->cmn_right;
203*fbc9049bSSascha Wildner 		} else
204*fbc9049bSSascha Wildner 			break;
205*fbc9049bSSascha Wildner 	}
206*fbc9049bSSascha Wildner 	l->cmn_right = t->cmn_left;
207*fbc9049bSSascha Wildner 	r->cmn_left = t->cmn_right;
208*fbc9049bSSascha Wildner 	t->cmn_left = N.cmn_right;
209*fbc9049bSSascha Wildner 	t->cmn_right = N.cmn_left;
210*fbc9049bSSascha Wildner 	return (t);
211*fbc9049bSSascha Wildner }
212